50 THE INTERFEROMETRY OF 



To enlarge the fringes, the prism P' may be rotated around a horizontal 

 axis parallel to LT. The fringes then also rotate, but the increase of size so 

 obtained is usually not striking. Moreover, no observable effect, either on 

 the size of fringes or on the range of displacement, is produced by inserting 

 compensators in one beam or both. If M and N are moved together toward 

 the right or left, the result is not appreciable. A great variety of different 

 adjustments showed a range of displacement, at M, about the same (0.06 cm.), 

 whether the patch of light on the prism was wide or narrow. The range of 

 fore-and-aft motion of P' within which fringes are visible was 0.52 cm. They 

 vanish quite abruptly when the light is near the edge of the prism, although 

 both spectra are still strongly visible. When the light is nearer the base of 

 the prism they vanish more gradually. Definite strips of white light on both 

 sides of the prism, therefore, cooperate to produce the fringes. The remainder 

 of the illumination is ineffective. The distance apart of c and c r , as modified 

 by fore-and-aft motion, curiously enough, is here without marked influence. 

 It is true, however, that the largest fringes were obtained when the two pencils 

 of light from M and N coincided at the objective of the telescope, although 

 the D lines were in this case far apart. The attempt to find a systematic 

 method for enlarging the fringes failed, possibly because the prism angles 

 were not quite identical. The striking contrast in the results obtained here 

 in comparison with those of the preceding paragraph, although both methods 

 are essentially the same, is noteworthy. 



It is for this reason that I thought it desirable to test the method in figure 3 1 , 

 which accomplishes with a prism what was done in my original experiments 

 with reversed spectra, by the aid of a grating. In the figure the incident 

 beam of white light L from a collimator strikes the 60 prism at its edge, and 

 is then refracted into the paired pencils a, a'. These are reflected normally 

 by the opaque mirrors M and N, again refracted by P as each pencil nearly 

 retraces its path. The return beams, however, are given a slightly upward 

 trend, so as to impinge on the opaque mirror m (curved or plane). The rays 

 reflected from m, in such a way as to avoid the prism P, may be reunited in 

 the focus F observed by the lens T, or (if parallel) collected by a telescope at T. 

 In view of the prism, the spectra are small and reversed, but may be brought 

 to overlap at the red ends, which are towards each other. 



The small dispersion makes it necessary to use a strong telescope if the 

 Fraunhofer lines are to be visible and the D lines separated. Usually the 

 two doublets will be at a small angle to each other, but this does not mar 

 the interferences. When the adjustment has been made symmetrically, 

 a strong linear phenomenon maybe found not differing in appearance from the 

 results obtained when a grating was used at P, figure 3 1 . When the mirror 

 M is displaced, however, the fringes appear in the form of multiple vertical 

 hair-lines, which grow coarser until but a single dark line flanked by a bright 

 line is visible. With further displacement the phenomenon again vanishes in 

 passing through multiple hair-lines. This appearance of hair-lines is one 

 distinguishing feature ; but a much more important result is the small range 



